Study of Traps in Special Doped Optical Fiber Radiation Sensors via Glow Curve Analysis

New perspectives are evolving in respect of the wide range of potential thermoluminescence dosimetry (TLD) applications of various forms of optical fiber [1,2]. As an instance, optical fiber TLDs has been shown to be effective for a range of photon sources, from visible and UV, through to xand gamma-rays of different energies [3-5]. Further studies have concerned silica fibers of various shape and doped-core dimension, flat optical fibers being shown to lead to improved TL response [6-10], it also being demonstrated that the TL response of fibers will vary depending upon core dopant concentration as well as core size [11-13].


Introduction
New perspectives are evolving in respect of the wide range of potential thermoluminescence dosimetry (TLD) applications of various forms of optical fiber [1,2]. As an instance, optical fiber TLDs has been shown to be effective for a range of photon sources, from visible and UV, through to x-and gamma-rays of different energies [3][4][5]. Further studies have concerned silica fibers of various shape and doped-core dimension, flat optical fibers being shown to lead to improved TL response [6][7][8][9][10], it also being demonstrated that the TL response of fibers will vary depending upon core dopant concentration as well as core size [11][12][13].
Although commercial TLDs have typically developed, following up upon the favourable outcome of studies of various constituents, for optical fibers the study of TL yield have been much more limited, most typically with germanium, aluminum, oxygen, phosphorus and nanomaterial cluster dopants [4,[14][15][16]. Present study seeks to help address this lack, expanding investigations to include additional rareearth dopants, examining TL response and associated kinetics analysis of such optical fibers.
The characterization of the defect centers forms a crucial step in understanding the mechanism of TL [17]. In this context, analysis of glow curves offers a sensitive and suitable technique for such study. Here, in present study, new materials are tested to examine their effect, both on TL response and kinetic parameters.

Optical fiber fabrication and preparation
The samples are standard circular cross-section optical fiber of 125 µm outer diameter, fabricated using the facilities of the University of Malaya fiber pulling lab. The 9 µm central core that forms the dopant channel has been confirmed using the energy-dispersive x-ray (EDX) facility of an SEM, doped-core and silica cladding mappings being obtained for all but one of the samples (Figure 1). For Al-doped fiber, EDX mapping of the core represents a severe challenge, due to the neighbouring atomic numbers of Al (Z=13) and Si (Z=14), providing for limited differential X-ray fluorescence production, further confounded by the associated low energy emissions (Figure 1b). The doped elements are compared with SiO 2 , BaF, GaP, Al 2 O 3 , TmF 3 as a reference for EDX identification. The characteristic peaks through which the dopant materials are detected are K α1 for O, Si, Al, Y, Ba and Ga and L α, β for Tm. Table 1 shows the concentration of each element present in the different samples.

Sensor Networks and Data Communications
The outer polymer cover to the fibers has been carefully removed using a chloroform solution, soaking the fibers in the solution for 30 seconds. The samples are subsequently cleaned with acetone and carefully cut to 5 mm lengths using a fiber cleaver. Two sets of five samples each were chosen from each type of fiber, the uncertainly in fiber length being ± 0.5 mm. Since the fiber samples have closely identical mass density, also being carefully selected for uniform length, mass normalization of the results has not been required in carrying out analysis and comparisons.

Irradiation and readout
Prior to any irradiation run, in order to remove any prior radiation history, all samples are annealed at 400°C for one hour, subsequently cooled to room temperature. The samples have then been exposed to a dose of 8 Gy dose delivered either by 6 MeV electrons or 6 MV photons, use being made of a Varian 2100 C linear accelerator. The samples were positioned on the surface of a solid-water TM phantom. The field size and surface to source distance were set at 20 × 20 cm 2 and 100 cm respectively. A square applicator with 20 × 20 cm 2 aperture size was used for electron irradiation.
The TL response of radiated samples was obtained using a Harshaw 3500 TLD reader, the irradiated samples being thermally stimulated from 50°C to 400°C, provided at a heating rate of 20°C/s. The generated glow curves have been analyzed using a second order kinetics model for TL deconvolution.

Results and Discussion
The differential TL responses of the five sets of optical fibers with respect to a dose of 8 Gy are illustrated in Figure 2 for both photon and electron irradiations. The TL yields are similar for both the photon and electron irradiations, with a deviation between them of less than 4% based on the mean values.
In Tables 2-6 show the first peak (referred to as Trap 1) to have an activation energy of 0.9-1.1 eV, a peak intensity of 2-2.9 μC, and an emission wavelength of between 1100 and 1400 nm pointing to independence from the dopant material, instead being more related to the substrate or the silica preform itself. The activation energy suggests association between O 2 and O 2defects in amorphous and fused silica [18].
Examining Figures 2-4, the numerous traps in Al-Y-Tm-doped and Ba-doped fiber provide for the greater response of these two forms over that of other samples, each of which are composed of lesser numbers of traps (Figures 5-7). This supports the underpinning basis of greater numbers of traps producing greater absorption. According to Figure  3 the combination of Al, Y and Tm as fiber co-dopants results in an increased number of trap centres and, as a result, enhanced absorption. Figure 4 for Ba-doped silica, provides for lower intensity glow peaks compared to that for Al-Y-Tm-doped fiber, albeit with similar energy traps formed in the two types of fiber.
In accord with Tables 2 and 3, the major difference between Al-Y-Tm-doped fiber and Ba-doped fiber is the intensity of the individual    peaks. In the first case, the more energetic traps have the greater capacity for storing memory of the irradiation. The complex shape of the Ba-doped fiber glow curve is posited to be due to deep interactions of this element with the host silica preform. Figures 5 and 6 show deconvolutions for optical fibers containing the same dopant (Al-Tm), one with a greater concentration of Al inside the fiber core than the other. Tables 4 and 5 show that in increasing this dopant concentration, there is no practical change in the peak positions (peak T ≈ 160-166 and 193-206 for traps 2 and 3 respectively), but the intensity is increased, the exception being for trap 1 which is independent of dopant as discussed earlier (peak I 1 =1.5 and 1.6 → peak I 2 =2.2 & 2.3 for traps 2 and 3 respectively).
The simple glow curve of Ga-doped fiber is shown in Figure 7. The        main TL is generated from the first glow peak which is from the silica preform, while lesser TL is contributed from peak 2, from Ga traps.

Conclusion
The similarity in activation energy of first glow peak for all fiber samples indicates the TL response of the substrate, regardless of the dopant in the fiber. It is concluded that different concentration of similar dopant will only affect the intensity of relevant glow peaks while other factors will remain the same. An interesting observation from this study is that co-doping of the fiber is associated with an increase in the TL yield, pointing to a need for optimization of the choice and levels in use of co-dopants.